Google launches Gemini 3.6 Flash and two specialized agent models

Google has introduced three new additions to its Gemini Flash lineup: Gemini 3.6 Flash, Gemini 3.5 Flash-Lite and Gemini 3.5 Flash Cyber. The releases are aimed at developers building production-scale AI agents, where latency, token usage and predictable performance can directly affect operating costs.

Gemini 3.6 Flash is positioned as the general-purpose workhorse. Google says it uses 17% fewer output tokens than 3.5 Flash in the Artificial Analysis Index, while requiring fewer reasoning steps and tool calls for multi-step tasks. It is also priced below its predecessor at $1.50 per million input tokens and $7.50 per million output tokens. Google reports gains in coding, computer use, multimodal analysis and knowledge work, including stronger results on the DeepSWE, MLE Bench, OSWorld-Verified and GDPval-AA v2 benchmarks.

For applications that prioritize speed and price, Gemini 3.5 Flash-Lite is described as the fastest and most economical model in the 3.5 family. The company cites an output rate of 350 tokens per second and says it improves on earlier Flash-Lite versions in agentic workflows.

Gemini 3.5 Flash Cyber takes a narrower approach. It combines a specialized cybersecurity model with Google’s CodeMender code-security agent, reflecting the company’s emphasis on pairing models with supporting agent infrastructure rather than treating the model as a standalone tool.

For AI builders, the releases expand the options for routing workloads: 3.6 Flash for stronger general agent performance, Flash-Lite for high-volume low-latency tasks, and Flash Cyber for security-focused workflows. Google also says Gemini 3.5 Pro is being tested with partners and that training has begun on Gemini 4.

Source: Google DeepMind Blog